Re: Managing the JIT

Armel wrote:
> "Philip Herron" <herron.philip@googlemail.com> a icrit dans le message de
>
>> [...]
>> something. I am having trouble finding more stuff on how this works
>> would be great if you could point me in the right direction?! :)
>>
>
> I just felt on the asmjit project : http://code.google.com/p/asmjit/
> it seems that it does not contain megatons of code, you may find
> interesting to read it.


Hey


Thanks for that link its pretty interesting i have been toying with my
own idea of creating a mini jit execution system i guess. I have a bit
of a mini x86 code-gen for basic things like additions multiplications
and arrays etc. But its by no means complete, and i dont think i ever
would, because LLVM would be much more complete than anything i could do
on my own. But i started with my interpreter to jit functions using llvm
not fully working but alot of the test code is there so when functions
are called:


main()
{
    do_x()
    do_y()
}


etc...As soon as a top-level function was called like do_x or do_y etc
it would go into my symbol table of the functions and iterate though and
jit any other lower level function calls down the tree as it needed but
i only went 3 functions deep was my test but it should work on as i just
used stacks to maintain my place in the tree. I am wondering how this
impacts object oriented languages.


If you have a function that creates an instance of a class do you jit
the data structure along with its member functions or do the member
functions as you called them. I guess i would just do it as i called
them for now. Then a non-deterministic language like haskell could get
confusing also. Hmm i think i need to do some experimentation. I'll be
posting links to my interpreter soon enough and my gcc internal
documentation soon. Its taking a lot longer than i thought to do.


--Phil
http://redbrain.co.uk

Return to the comp.compilers page.
Search the comp.compilers archives again.